NASA Aeronautics Research Mission Directorate (ARMD)

2015 Seedling Phase II Technical Seminar

November 17 & 19, 2015

Concept Demonstration of Dopant Selective Reactive Etching (DSRIE) in

Silicon Carbide Robert S. Okojie, PhD

NASA Glenn Research Center, Cleveland OH

Email: robert.s.okojie@nasa.gov; phone: (216)433-6522

https://ntrs.nasa.gov/search.jsp?R=20160009223 2020-03-23T14:26:17+00:00Z

Outline

Motivation

Technical Challenge

Observed Phenomenon

Technical Approach

Results

Innovation Impact on Mission if implemented

Conclusion

Problem Statement-Motivation

Lower emissions (LE) jet-engines are critically dependent on lean-burning (LB) operation (low fuel/air ratio).

LB combustion is susceptible to thermo-acoustic instabilities, producing pressure oscillations that can reduce component life and potential engine failure.

Existing CFD instability prediction models have high uncertainties in high temperature regime.

SoA dynamic pressure sensors are placed distance away from test article-limits frequency bandwidth; Water-cooled sensors introduce “vortex noise” that corrupt signal.

Need high temperature (>500 C) pressure sensors to validate CFD models.

HeatRelease

Oscillations

AcousticOscillations

Motivation (SoA dynamic pressure sensor versus SiC Static Sensor)

Flush mounted SiC dynamic pressure transducer

PCB dynamic pressure transducers mounted on

semi-infinite line

0

0.1

0.2

0.3

0.4

0.5

0 200 400 600 800 1000 1200 1400

p

(p

si, p

-p)

Freq. (Hz)

26.06.2013 15-11-19 (703 F) 26.06.2013 15-11-50 (703 F) 26.06.2013 15-12-29 (703 F)

26.06.2013 15-46-49 (703 F) 26.06.2013 18-33-07 (1085 F) 26.06.2013 18-34-11 (1086 F)

PCB Dynamic pressure transducer at P3

0

0.1

0.2

0.3

0.4

0.5

0 200 400 600 800 1000 1200 1400

p

(p

si, p

-p)

Freq. (Hz)

26.06.2013 15-11-19 (703 F) 26.06.2013 15-11-50 (703 F) 26.06.2013 15-12-29 (703 F)

26.06.2013 15-46-49 (703 F) 26.06.2013 18-33-07 (1085 F) 26.06.2013 18-34-11 (1086 F)

SiC 235 static pressure transducer at P3

161 Hz (1085 F, 1086 F)

0

0.2

0.4

0

0.2

0.4

p

si, p

-p

psi

, p-p

Uncooled SiC static pressure sensor at P3

Water cooled PCB dynamic pressure sensor at P3

Results:

• Reduced SiC data compared favorably well with the reduced PCB data.

• Low frequency noise (<500 Hz) suppressed the static SiC sensor output.

Value PropositionA real SiC dynamic pressure sensor will operate uncooled

at high temperature, increase sensitivity, increase S/N ratio at lower frequencies

Note: The thick diaphragm SiC sensor not designed for dynamic operation

Technical Approach

Based on circular plate theory, maximum deflection (for deflections << thickness) of a clamped circular diaphragm is expressed as:

Where: d=maximum deflection (m); P=applied pressure (Pa); r=diaphragm radius (m); E=Young’s Modulus (Pa); h=diaphragm thickness (m); =Poisson ratio

E = 475 GPa for SiC; h (research goal) = 5 µm; r =1 mm; = 0.212Calculated maximum stresses on diaphragm: Center=6.27 MPa; Clamped

edge=-10 MPa. Both well below the fracture strength of SiC

The predicted applied pressure is ~345 Pa (0.05 psi)

d=𝟑𝑷𝒓𝟐

𝟏𝟔𝑬𝒉𝟑(𝟏 − 𝟑)

Goal: Obtain SiC diaphragm thickness 5 µm or less!

Goal: Realize ultrathin diaphragms in 4H-SiC

Reactive Ion Etching (RIE) for Diaphragm Fabrication: The Basics

Halogen-based gas (i.e., SF6) is ionized to create F ions and neutrals. Neutrals react chemically with target low etch rates.

With argon added and ionized, Ar+

physical bombardment of target increases surface area increased chemical reaction and higher etch rates.

E-Field

F-F

-F-F

-F

-Neutrals

Conventional SiC RIE

Conventional RIE (with SF6 + Ar) of semi-

insulated (SI) 4H-SiC substrate to create

thinner diaphragms resulted partial

etch-through of the micro diaphragms.

Observed free standing, 2-µm thick

“cantilevers” of the patterned highly

doped n-type homoepitaxially grown

4H-SiC piezoresistor layer on the SI SiC.

Etch selectivity between SI and highly

doped n-type SiC is proposed.

If controlled and reproducible, could

lead to the realization of ultrathin

diaphragms by RIE.

Hole

Etched diaphragm in semi-insulating SiC

Boss

Free standing, 2 m highly doped

n-type SiC resistor

Semi-Insulating

SiC substrate

Hole due to over-etch

Free standing, 2 m highly doped n-type

SiC resistor

RIE

Experiment: RIE Etching of SI and N-type 4H-SiC with SF6

SF6= Flow Rate (sccm) SI N-Type(Nd=3.8 x 1019 cm-3)45 3 860 4 975 5 10

Calibration patterns etched in n-type and semi-

insulating SiC samples.

Fixed process conditions: RF Power=400 W; Base Pressure = 25 mT; Etch Time=2 hours

Photolithography and Al pattern definition with

calibration mask

RIE

Al mask stripped off and performed depth profile

measurement

DepthProfile

Etching selectivity between SI and doped n-type substrates in SF6 gas only.

Selectivity Between SI and N-Type 4H-SiC

480

500

520

540

560

580

30 40 50 60 70 80

Etc

h R

ate

(Å/m

in)

SF6 Flow Rate (sccm)

SI-SiC N-Type SiC (3.8 x 10^19 cm^-3)

RF Power = 400 WBase Pressure = 25 mT

Etching selectivity between highly doped p- and n-type 4H-SiC substrates in SF6 gas only.

600

650

700

750

800

850

900

950

1000

30 40 50 60 70 80 90

Etc

h R

ate

(Å/m

in)

SF6 Flow Rate (sccm)

DH0852-14 (p-2e20), 0.5 µm epi BQ0339-10 (n>2e19), 2 µm epi

RF Power = 400 WBase Pressure = 25 mT

Selectivity Between N- and P-Type 4H-SiC

DSRIE Release of N-Type Cantilevers from SI SubstrateFirst proof of concept

11

Suspended 2 µm

thick single crystal

SiC piezoresistor

Proof mass

suspended by 2 µm

thick single crystal

SiC piezoresistor

SEM paste!

Suspended 2 µm

thick single crystal

SiC piezoresistor

Investigation of Selected HalidesBCl3, HBr, and Cl2

• Wafer Clean10 minutes in Acetone and 10 minutes in IPA

• Nickel Deposition230 nm thick nickel in Cooke E-beam Evaporator

• Optical lithographyAZ 5214 E 4000 RPM about 1.3 um thick; Bake at 110 degrees for 1 minute; Expose with NASA mask for 40 seconds at Karl Suss MJB Aligner; Develop in AZ327 for 4 minutes.

• Wet Etch4 minutes in Nickel Etchant (Type TFB from Transene Company); 10 minutes in Acetone to remove photoresists

• ICP-RIE Etch100 W for RIE and ICP power settings; Gas flow rates from 5 sccm, 10 sccm, 15 sccm, to the maximum of 20 sccm; First depth measurement with Alpha Step

• Nickel Removal4 minutes in Nickel Etchant; Second depth measurement

Etching with HBr

80

90

100

110

120

0 5 10 15 20 25

Etc

h R

ate

(Å/m

in)

Flow Rate (sccm)

Semi-Insulating p-type (2x10^19 cm^-3) n-type (3x10^19 cm^-3)

HBr Only ICP Power=100 W

Etching with HBr + Ar

0

50

100

150

200

250

0 5 10 15 20 25

Etc

h R

ate

(Å/m

in)

HBr Flow Rate (sccm)

Semi-Insulating p-type (2x10^19 cm^-3) n-type (3x10^19 cm^-3)

HBr + ArAr fixed at 20 sccm

Etching with BCl3

60

85

110

135

0 5 10 15 20 25

Etc

h R

ate

(Å/m

in)

Flow Rate (sccm)Semi-Insulating p-type (2x10^19 cm^-3) n-type (3x10^19 cm^-3)

BCl3 Only

Etching with BCl3 + Cl2 + Ar

100

150

200

250

300

0 5 10 15 20 25

Etc

h R

ate

(Å/m

in)

BCl3 Flow Rate (sccm)

Semi-Insulating p-type (2x10^19 cm^-3) n-type (3x10^19 cm^-3)

BCl3+Cl2 + ArCl2 fixed at 5 sccmAr fixed at 20 sccm

Summary of ER Selectivity Between SI and N-Type 4H-SiC

Gases Power (W) Flow rate (sccm)

ER

SF6 400 60 1.03

HBr 100 10-20 ~1

BCl3 100 15-20 1.33

BCl3 + Cl2+ Ar 100 BCl3 (15), Cl2

(5), Ar (20)1.37

Facility limitation prevented investigation of higher power and flow rates

Ultrathin Diaphragm with BCl3 + Cl2 + Ar etching

Full diaphragms

Cut out diaphragms

Wafer edge

3 µm 4HSiC n-type epilayer

150 µm 4HSiC SI substrate

GoalStart with fast ER recipe through most of SI substrate, then switch to DSRIE:BCl3 (15 sccm)Cl2 (5 sccm), Ar (20 sccm)Power 100 W

Results

Microtrenching

Ultrathin Diaphragm with BCl3 + Cl2 + Ar etching

Diaphragm thickness between 2 and 3.5 µm

Ultrathin Diaphragm with BCl3 + Cl2 + Ar etching

Observed microtrenching and etch non-uniformity across wafer causes reduced

yield. Etch conditions and chamber optimization are required to increase yield

Semi-insulated 4H-SiC substrate (300 µm)

Highly doped N-type 4H-SiC epitaxial layer (2 µm) Piezoresistor fabrication

e) Diaphragm formation: fast reactive ion etchingf) Nickel strip and switch to DSRIE to form

diaphragm

Contact metallization

a) Starting wafer b) Define, pattern, and dry etch piezoresistors

Highly resistivity polycrystalline CVD SiC (> 2 µm)

d) Nickel electroplated maskc) Chemical vapor deposited SiC

g) Piezoresistor oxide passivation and contact

via etch,

h) Metal deposition, pattern and etch

Fabrication Process Steps of Ultrathin Diaphragm Sensors Using DSRIE

Passivation and Contact via etch

~7 µm

Fabrication Process Steps of Ultrathin Diaphragm Sensors Using DSRIE

Diaphragm microtrench eliminated

Diaphragm thickness between 4 and 7 µm

Before

Fabrication Process Steps of Ultrathin Diaphragms-Sensor Side

Piezoresistor

Contact via

Outline of diaphragm

Next Step: Complete fabrication, package and test SiC dynamic pressure

sensor with ultrathin diaphragms

Innovation and Impact on Mission

Mission Challenges: Thermoacoustic instabilities in combustors are known to be precursor to flame-out or damage to engine components.

Existing instability prediction models have high uncertainty margins. Need environmentally robust and reliable pressure sensors for model validation and improvement.

Existing Technology Gap: SoA sensors placed feet from test article-limits frequency bandwidth; Water cooling adds “vortex noise” to corrupt signal. Robust and reliable sensors needed for direct (no water cooling) measurement of sub-psi dynamics at >500 ºC.

600 ºC SiC pressure sensor technology currently exists, but SiC fabrication technology cannot produce the ultra-thin diaphragms to achieve the high sensitivity needed to resolve sub-psi pressure dynamics.

Innovation: DSRIE will result in ultra-thin (< 5 µm) SiC

diaphragms to accurately resolve sub-psi dynamics at temperatures in excess of 500 ºC (current capability).

Flush mounted

SiC dynamic sensor

Benefit of SiC Etch Stop Mechanism

~5 µm etch stop diaphragm

Semi Insulating SiC

Summary/Conclusion

We investigated the possible existence of dopant and conductivityselectivity in SiC during reactive ion etching with selected halidegases;

Within the limits of experimental space, the maximum etchselectivity between the SI and n-type SiC was 1.37, enough torelease free standing n-type cantilevers and diaphragms of less than10 µm thick.

Applied process to fabricate array of ultrathin SiC diaphragms forpressure sensors.

Acknowledgement

Sponsor: ARMD Seedling FundUniversity of Illinois Urbana Champagne-Matrix Experiments

Dr. Robert Shinavski (Roll-Royce)- SiC deposition Kimala Laster/Ariana Miller - Sample fabrication

Beth Osborn/Michelle Mrdenovich-Sample preparationLaura Evans and Pete Bonacuse - SEM images

Kelley Moses and David Spry- RIE system health maintenance